Stereo Generator Page

A Digital Stereo Encoder.

A piece of equipment commonly used in FM Broadcast intallations to produce a MPX (multiplex) signal to modulate the broadcast transmitter carrier frequency. This MPX signal contains the information required to transport the left/right channel and allow the receiver to decode the signal at the receiver in to the respective left/right channels. The design given here is based upon a DSP (digital signal processor). Using this approach it is possible to achieve a near perfect MPX signal. In fact with a little effort the MPX specification can exceed any practical receiver specification.

The stereo maths

The stereo generator block diagram

The stereo frequency spectrum

Choice of components

Here the various tradeoff's involved in the choice of components is discussed. Of course, the first requirement is to settle on a specification.

For example:

Is the input digital or analogue or both?

Is the output analogue or digital or ?

Do we include RDS or other sub-carriers.


Choosing components

The software

The software depends upon the hardware and the hardware depends upon the software philosophy!!!! That is to say the software approach will influence the components choosen and the way these components interact (read connect up). It is a chicken and egg situation that requires the project to be invested in a single mind or a very closely controlled/bound team. The solution applied here uses a decoupled software approach. The hardware decides when the input sample is taken (this is a DSP, remember). The DSP (application software) then toddles off, processeses the information, produces a result and stores the result (this is important). The hardware then recovers the result from the DSP when required in order to generate the appropiate output signal. This means the hardware operates as a synchronous system, whereas the DSP functions in an asynchronous mode. Provided the DSP does ALL that is required in the time allocates by the hardware between the taking of the input sample and the request for the output sample there is no problem. Using this approach the DSP processesing time can and will vary.

For example:

The processing time will vary according to the expansion/compression/auto gain control settings

The source of the input, analogue or digital have different pre-processing requirements.

The addition of RDS substantially increases the processing time.


Designing the software

Integration and tidying up

So far the digital core has been talked about. There are still a number aspects relating to the analogue processing of the signals. To a much lesser extent the digital signals have to be "prepared".

For example:

Elimination of the spurious MPX signals resulting from the digital to analogue conversion.

Correction of the imperfect digital to analogue conversion

Protection of the inputs from overrange signal levels.

Provision for different exciter/radio link drive levels and injection of external sources (read RDS or SCA)


Finally, there is a much neglected feature: the power supply. A perfect analogue/digital design can easily be let down by a bad power supply. Leaving the power supply to the last minute when production time schedules loom can and WILL result in a sub-optimal if not downright bad design.

Result 1: the hardware designer will try to correct the noise, ramp-up, stability problems by adding more hardware (read decoupling, filters etce...).

Result 2: the software designer will try to add more processing (read filter the samples, smooth out glitches etc...).

So fix the supply format at the same time as the hardware is set and use that format ASAP during the development.

Integration and other bits